Ssh Manager MCP Server

SSH remote server management for AI assistants to execute commands and sync files.

Data last scanned yesterday · Reviewed 3 weeks ago

53 1 dangerous tool4 tools★ 431JavaScriptMIT

Overview

SSH Manager MCP Server allows your AI models to interact directly with remote server infrastructure. By integrating with Claude Code and OpenAI Codex, it enables users to run shell commands, manage database operations, perform file transfers, and automate complex DevOps workflows across multiple SSH hosts. The server features advanced configuration management, allowing users to define server groups within their environment files or TOML configurations. It also includes cross-platform support for synchronization tasks, including robust handling of local filesystem paths on Windows.

Our verdict

This project carries a high-risk rating with a security score of 48/100. Because it remains community-maintained and has not undergone an editorial security review, users should be cautious. The lack of verified authentication and read-only mode protocols means AI models may have broader system access than intended; this tool is best suited for experienced developers working in sandboxed or isolated development environments.

Active development with a recent commit history
Small dependency footprint with only five direct packages
Supports grouping servers directly in configuration files
Open-source MIT license permits modification

Authentication methods remain unreviewed
Read-only operational modes have not been validated
Not maintained by an official vendor organization

Setup

Claude Desktop

{
  "mcpServers": {
    "mcp-ssh-manager": {
      "args": [
        "-y",
        "mcp-ssh-manager"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
mcp__ssh-manager__ssh_execute	Execute a command on an SSH server.	high
mcp__ssh-manager__ssh_upload	Upload a file to an SSH server.	medium
mcp__ssh-manager__ssh_download	Download a file from an SSH server.	low
mcp__ssh-manager__ssh_sync	Synchronize files between the local machine and an SSH server.	medium

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›Can I categorize my SSH servers into groups?

Yes, you can assign an optional group to each server directly in your .env or TOML configuration file, such as tagging a server with group = "production".

›How do I use my local SSH keys on a remote host without copying them?

You can enable the opt-in FORWARD_AGENT option per server, which functions like OpenSSH's ForwardAgent yes and uses keys from your local ssh-agent.

›Does this server support Windows hosts?

Yes, the ssh_sync functionality supports Windows, including automatic conversion of drive-letter and UNC paths for rsync operations.

›What happens if I try to create an SSH tunnel on an already-bound port?

The server will return a normal error instead of crashing the entire process, ensuring that other pooled SSH connections remain active.

›Are database helper commands secure?

Yes, all ssh_db_* arguments are shell-quoted to prevent command injection, ensuring that caller-controlled values are safely handled.